DDraceNetwork - developer

@Voxel what are you currently trying?

00:31

you could make a "team killed" message that shows the whole team in the kill feed

huh

00:33

I'm just basing the logic from teams.cpp, in CGameTeams::OnCharacterDeath()

00:33

line 1056: if(GetTeamState(Team) != TEAMSTATE_OPEN)

heinrich5991

you could make a "team killed" message that shows the whole team in the kill feed

Wait, make a netmsg for that? I kind of did that already

but why do you need the team state then?

00:36

if the server sends "whole team was killed" then the whole team was killed

And how do we determine that?

that the whole team was killed?

00:38

that happens in server code

Yeah

00:39

Exactly

you're asking me how the server determines whether the whole team is killed?

Not really, I'm asking how it knows when to kill off the entire team when someone dies, or just that person

00:40

Which is why I was wondering about teamstates

the server already does that

00:41

so you could go into the code and add the "the whole team was killed" message into the server code there

00:42

(and find a way to suppress the other individual kill messages)

But how do I fluently communicate that to the client-sided code, which you said can't be linked with server sided code? (edited)

no client side code was involved in my description so far

00:44

my client side code would be 'if I get the "the whole team is killed" message, then display that the whole team was killed' (edited)

Did you see my update to #6437 yet

https://github.com/ddnet/ddnet/pull/6437

[DRAFT] Make team kills condensed in killfeed by VoxelDoesCode · Pu...

For those who still have the killfeed on, I wanted to modernize it, by having team kills become one line instead of a bunch! This is still in progress, because there are some technicality issues I...

I'm just clang formatting it right now

I think that's a bit complicated right now

00:56

you're using two messages to transport information about one kill

00:56

NETMSGTYPE_SV_KILLMSG → one tee was killed

00:56

NETMSGTYPE_SV_KILLTEAM → a whole team was killed

00:57

no need to reimplement team state logic in the client this way

00:57

if a team is killed, only send one NETMSG_SV_KILLTEAM message

00:57

no NETMSG_SV_KILLMSG

00:57

if someone from a team dies without killing their entire team, only send a NETMSG_SV_KILLMSG

00:58

then the client code will be very easy

oh

01:16

i'll try and refactor all of this tommorow

01:18

Does this mean I have to m_aKillmsgs[m_KillmsgCurrent] = Kill; for both messages in killmessage.cpp?

yes

I'll refactor everything tommorow, but in its current state it produces the exact output I want it to

cool!

01:30

I really like the idea, but I think I said that already

Thanks! (edited)

02:10

Actually, I can split it right now, it shouldn't be too hard.

@heinrich5991 Okay, I submitted another PR. Hopefully I understood you right.

03:04

I'll be squashing it with the other one when I know it's perfect.

now you're sending one NETMSG_SV_KILLTEAM per team member,I think

03:56

instead of just one for the whole team

03:56

if I understand the code correctly

heinrich5991

now you're sending one NETMSG_SV_KILLTEAM per team member,I think

void CCharacter::Die(int Killer, int Weapon, bool Single) has a new bool (Single) for if it's only going to be one tee dying By default it's true, but if it's false the killmessage packet doesn't even send (edited)

ah, I see

04:14

yea, the logic looks good now, then

@Voxel Change Parameter Single to "Teamkill" or "SendKillMsg"

06:27

and send or not send the kill msg according to that

06:27

And pass it from where the team kills each character.

06:27

And send a team kill msg before then

bors[bot]

[ddnet:master] 2 new commits

f15d975 Update brazilian_portuguese.txt - rffontenelle 15e83dd Merge #6473 - bors[bot]

07:09

bors[bot]

[ddnet/ddnet] Pull request closed: #6473 Update brazilian_portuguese.txt

Marmare314

[ddnet/ddnet] Pull request opened: #6474 update selected group correctly after dragging

When dragging a layer up, the wrong layer is selected after. Should be merged before #6467.

Checklist

[x] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test (especially base/) or added coverage to integration test
[ ] Considered possible null pointers and out of bounds array indexing
[x] Changed no physics that affect existing maps
[ ] Tested the c...

bors[bot]

[ddnet:master] 2 new commits

927c801 update selected layer correctly after dragging - Marmare314 b31d28d Merge #6474 - bors[bot]

09:43

bors[bot]

[ddnet/ddnet] Pull request closed: #6474 update selected group correctly after dragging

is github down

12:27

haha

12:27

@deen HEH???

12:59

@Ryozuki wtf this is so unexpected

12:59

LMAO

12:59

you can create a Con folder without error

12:59

wtf

13:00

needless to say what version is this lol

mrpg

1

we hacked windows

why do u ping everyone

you coded it didn't you kek

13:11

Does windows break if you have a Con folder?

tf are u talking about

13:13

xd

as you can see u broke your setup. a con directory now exists and u have to reinstall windows

no disrespect but sometimes i dont understand you xD

i correct myself, windows is now broken for you forever. please install linux to continue

1

fokkonaut

And pass it from where the team kills each character.

wait, don't i already do that? im confused i just woke up

ryo, i was joking (edited)

13:23

and about the Con folder, I was genuine xd

Voxel

wait, don't i already do that? im confused i just woke up

didnt Check

Jupstar ✪

i correct myself, windows is now broken for you forever. please install linux to continue

isn't it more than Con?

13:23

some random hash id (edited)

13:24

in curled brackets

{}

Con was DOS times I think used for the I/O of hardware

13:24

shouldn't be like that anymore

13:25

but maybe I am clueless

i managed to get a git push through

13:25

default

Con was DOS times I think used for the I/O of hardware

con still can't be a file name on modern windows

a file name

13:34

but in this case it was a folder

Cellegen

and about the Con folder, I was genuine xd

ah, my bad, i didnt know Con name was special or meaningful

13:40

so to me i was, wtf is this guy saying lol

13:40

xD

Windows doesn't allow you to do that, so it was like dark magic for me xd

@Chairn what do you think about https://github.com/chipsalliance/chisel

GitHub - chipsalliance/chisel: Chisel: A Modern Hardware Design Lan...

Chisel: A Modern Hardware Design Language. Contribute to chipsalliance/chisel development by creating an account on GitHub.

13:46

https://www.chisel-lang.org/

Chisel/FIRRTL: Home

Chisel/FIRRTL Hardware Compiler Framework

Cellegen

@deen HEH???

Doesn't work for me on Windows 10. Are you using Windows 11? Maybe it was changed there.

Ye

13:49

Same issue xd

13:50

eh, you mean the folder creating part? maybe it's changed on windows 11

13:50

but windows 11 doesn't allow you still

Maybe it's not allowed when you use Windows Explorer to create the folder but it works with the Windows API now, so they game can create the folder

so all i need to do is to rename Single to KillMsg and then that's it? Idk what i'm supposed to be moving where @fokkonaut

default

but in this case it was a folder

I'm pretty sure folder names aren't different in that respect

heinrich5991

I'm pretty sure folder names aren't different in that respect

yeah, I can't create a "con" or "Con" folder in the game on Windows 10

13:54

Must be a change in Windows 11 or a different option in Windows to allow some of the previously forbidden names

either that, or im just that talented

you can create a con folder

13:56

with \\?\C:\path\to\con

13:56

but you need the other syntax which ddnet doesn't use

Yeah, but we never use long paths internally

13:57

This also allows going beyond the normal maximum path length

yup

13:57

the normal maximum path length is super weird in windows ^^

default

but in this case it was a folder

u can't do a folder too

but wasnt there an option to change that

13:59

maybe its active in win11

13:59

aswell as the con stuff

Robyt3

Must be a change in Windows 11 or a different option in Windows to allow some of the previously forbidden names

w11 doesn't allow

ReiTW

w11 doesn't allow

Also ingame? Maybe Windows Explorer is different

ah ingame didn't try (not home) (edited)

you can try in python as well @ReiTW

14:00

if you have that on your work computer

heinrich5991

you can try in python as well @ReiTW

well as simple with os.makedirs("Con"), it doesn't work (edited)

14:10

oh lol fullpath works, but con only doesn't

Ryozuki

@Chairn what do you think about https://github.com/chipsalliance/chisel

looks way too verbose. I already knew it existed, but never really delved into it. Functionnal paradigm to describe circuit can be interesting, but not sure if it is the best (though still better than VHDL or Verilog i guess)

isnt verbose kinda good sometimes tho

14:38

depends on what kind of verbosity i guess

14:38

i like rust verbosity but not java one

Verbosity in a language is imo fine as long as you have tools to abstract it away

14:42

Though I'm not a HDL user, so idk if my opinion applies to those aswell

it also depends on your goals. sometimes making a language to explicit can make prototyping harder

HDL languages, especially VHDL are really verbose, yet they don't say anything about how the language is interpreted

14:44

it's up to the compiler to guess what you meant, it's not really standardized

Ew

14:45

SPICE also had this, what is it with you hardware people and your aversion to standards?

hardware people are driven by industry de facto...

14:47

you can't really do open hardware like you can do software open source

14:47

now, it's starting to be more open

14:47

but all the tools are proprietary ones

@Chairn have u used such languages in RL?

VHDL for teaching in Europe and some projects, SystemVerilog for my PhD

nice

right now, im preparing research and teaching projects to apply for Associate Professor position in France, and i have 71h10 mins left...

this community just epic

14:51

we have everything

Chairn

right now, im preparing research and teaching projects to apply for Associate Professor position in France, and i have 71h10 mins left...

Ask chatgpt for help

and i wrote less than 3 pages (but these 3 pages are common to all positions)

Learath2

Ask chatgpt for help

Never, we are thinking of banning it in my current school

with ryo now even a compiler guy

we already caught some students with it

Chairn

Never, we are thinking of banning it in my current school

In all seriousness. I agree. It has no place in academia

Chairn

we already caught some students with it

as long as the information is checked and correct it sounds like a useful tool to use

14:52

why ban it?

I don't care if you use it at your job and fuck over your boss, but if you want to learn/teach, it really does nothing but harm

Jupstar ✪

as long as the information is checked and correct it sounds like a useful tool to use

because they don't check the information...

14:53

for the same reason Wikipedia was said to be a bad tool when it appeared and students used it

i also did my english homework with google translate xD

they just copy paste, no critical thinking at all

yeah

14:53

but students simply are like that. for math its useless anyway

you can ask simple questions to students on what they "wrote", they will try to bullshit some stuff

Jupstar ✪

i also did my english homework with google translate xD

I think this is just a disservice to yourself. Unless you are far beyond the curriculum and you truly have nothing to learn

and for english its better than most teachers ez

Learath2

I think this is just a disservice to yourself. Unless you are far beyond the curriculum and you truly have nothing to learn

well it saved my sleep

14:54

i'd call it a success

Well is it not more of a time management issue? I'm sure you could have made some time for an english homework especially if it's below your level

14:55

There should be no need to sacrifice your sleep

nah was not possible for me

14:56

bcs i had friends and hobbies

Chairn

for the same reason Wikipedia was said to be a bad tool when it appeared and students used it

all tools can be used wrongly

14:56

wikipedia is amazing, and good pages have references to the sources

14:56

so u also read sources

14:57

students are a bit idiot and dont actually read the references xd

Jupstar ✪

bcs i had friends and hobbies

I find it hard to believe you didn't have 30 minutes for an english homework in an entire week

yeah i agree, but chatGPT pushes things a little bit too far and too fast imo

14:58

we'll need time to adapt

But anyway, you do you. If it was below your level there wasnt anything for you to learn, so not much harm

15 years, everyone will be using it

thats too long timespan

14:58

maybe yes back in 2010

14:58

but now things move faster

you can't accelerate social acceptation

i think u can

Learath2

I find it hard to believe you didn't have 30 minutes for an english homework in an entire week

why in an entire week? we got homework every day

14:59

and school until 16 o'clock

Jupstar ✪

why in an entire week? we got homework every day

You got homework for every subject every day?

Learath2

You got homework for every subject every day?

pretty much yes

We usually had one homework per subject per week

well seems like its different in ger

universe broken

Learath2

You got homework for every subject every day?

kinda ye

Huh, so you had every subject every day? How long was school even?

not but we got homework for each subject we got each day

i had most subjects 3 days in a 5 day week

usually just 1 day in between

15:01

but all kids prefer to play games duh

15:01

so u do it last day rushed

Jupstar ✪

i had most subjects 3 days in a 5 day week

Okay that sounds about right. And every time you got homework?

kek

Learath2

Okay that sounds about right. And every time you got homework?

especially in the earlier classes i'd say yes

Huh, we usually got a homework only on the last occurance of a subject within a week and we'd be expected to finish it until the next week first occurance

15:03

So that gave about 5, 6 days to do it

15:04

Different countries do handle it different apparently, no idea why I was surprised

Jupstar ✪

and school until 16 o'clock

yeah this sucks xd, had this up from fifth grade to tenth

I think 8 to 16 school is pretty universal except nordic countries

u have to be taught early that u a slave

they allowed us to prepare to exams this year

15:06

so no homework at all

It took me so long but I finally understand why they give homework. It's really where you learn things

not if u hate them

15:07

and thats the whole problem

15:08

if u like stuff u open to learn it

nah, homework is just method of controlling your learning curve

The problem is that kids mostly don't give 3 shits about learning anything

i am too hipster for our school form

im still learning but it became easier this year

gotta go take my train to go back to Paris, see you guys

the only good skill i took from my math homework is vieta theorem

Chairn

gotta go take my train to go back to Paris, see you guys

хорошего пути!

Until university honestly if you are smart-ish you can get away with not practicing at all. But when you get to uni and have no practice habit you finally understand why they were trying to get you to build a homework habit

Learath2

Until university honestly if you are smart-ish you can get away with not practicing at all. But when you get to uni and have no practice habit you finally understand why they were trying to get you to build a homework habit

i think this depends on university

15:10

and teachers

gerdoe

i think this depends on university

If you are planning to do anything numerical, unless you are a genius you'll need practice

ah yep xd

15:12

friends told me about useless subjects related to IT

15:12

not math

15:13

math is always good and addictive

Programming especially benefits from practice imo. The more you code the better you get

15:14

That's why I think chatgpt and copying from stackoverflow is not okay for students

Yo, why is the random skin button only cycles through the vanilla skins?

Also why I think C is a good language to get students started on

c is just good language

bors[bot]

[ddnet:master] 7 new commits

0ce70f8 add proof mode for menu backgrounds closes #3009 - Marmare314 9537656 use array again and consider checkpoints - Marmare314 8e2869e add tooltips - Marmare314 0cbc725 fix formatting and include array - Marmare314 7f0c052 fix formatting - Marmare314 93d204f show duplicate positions in tooltip - Marmare314 a27dbb1 Merge #6435 - bors[bot]

15:14

bors[bot]

[ddnet/ddnet] Issue closed: #3009 Editor should show standard locations for themes

15:14

bors[bot]

[ddnet/ddnet] Pull request closed: #6435 add proof mode for menu backgrounds closes #3009

Cellegen

Yo, why is the random skin button only cycles through the vanilla skins?

Hm, I guess only because it's coded that way and no one thought about it

It the cycling done through an Array of sort?

Learath2

Also why I think C is a good language to get students started on

the only thing c missing is ecosystem

15:16

thats why rust is better i think

15:16

but borrow checker uh

Yeah cargo is godsent

15:17

I don't think rust is a good language to teach beginners. Only because borrow checker is such an alien concept and I would want to stop them from using collections but that'd be extremely unidiomatic rust

15:17

Why teach them bad crippled rust?

rust and c exclusive and thats it

C is so small that it doesn't take long to understand the basics

If I was creating a software engineering curriculum I think I'd teach C and follow it up with C++ for teaching about OOP. I'd like to teach rust too but honestly 3 languages in a 3/4 year degree sounds annoying to me

no no no pls don't use cpp for teaching

gerdoe

no no no pls don't use cpp for teaching

15:21

why not

gerdoe

no no no pls don't use cpp for teaching

I have to show OOP and I'd rather not go with Java

cpp hides many things from programmer

Learath2

I have to show OOP and I'd rather not go with Java

c#?

15:22

same problem thouhj

15:22

though

gerdoe

cpp hides many things from programmer

This is what the C is for

15:26

I have some good ideas on how to produce a good software engineer. They should gimme a university

i'd probably teach c++98

15:27

and assembler

To introduce OOP? I had a similar idea in mind though I was thinking C++03

and then combine both

Since when is University for producing software engineers? Computer science has little to do with learning how to program, I'm afraid.

Learath2

To introduce OOP? I had a similar idea in mind though I was thinking C++03

yes and i think the best u can do is how OOP is "implemented" in assembly then

15:28

getting a feeling for vtables, memory layout etc

Robyt3

Since when is University for producing software engineers? Computer science has little to do with learning how to program, I'm afraid.

I said a software engineering curriculum. I didn't say a CS curriculum

imagine ddnet written in pure assembly (edited)

what even is pure assembly?

15:29

hardware binary code?

15:29

that can be displayed as e.g. x86 assembly

Voxel

imagine ddnet written in pure assembly (edited)

The extreme focus in theory is why the industry is plagued by people that are "programmers"

15:32

Mfw can't code without internet access pepeW

Voxel

imagine ddnet written in pure assembly (edited)

Would probably be pretty shit, modern compilers are far beyond what we can produce by hand nowadays

Learath2

The extreme focus in theory is why the industry is plagued by people that are "programmers"

many ppl that study CS dont want to "actually" be CS anyway

15:34

so i agree

15:34

u kinda have to be bit nerdy

Most of them effectively work as software engineers. Why not train them properly for it?

Is there any reason why the naming conventions are not checked with clang-tidy? I was thinking about trying to add that

ReiTW

C is so small that it doesn't take long to understand the basics

the university lecturers don't really understand C

16:15

with rust it seems to be a little bit better

16:15

because there's less UB

16:15

UB seems to be a hard concept to teach

why tho

marmare_314

Is there any reason why the naming conventions are not checked with clang-tidy? I was thinking about trying to add that

you can try doing that, however there's >1000 naming violations that I found when I wrote the script

its literally unexpected behavior xd

16:17

its called undefined but actually mostly ends up with unexpected

marmare_314

Is there any reason why the naming conventions are not checked with clang-tidy? I was thinking about trying to add that

those would need to be fixed, or there needs to be a way for it to only affect patches

gerdoe

its called undefined but actually mostly ends up with unexpected

that's the misunderstanding that people have with it, basically

heinrich5991

you can try doing that, however there's >1000 naming violations that I found when I wrote the script

I think clang-tidy supports changing them automatically but ill have to check

they think int function() { g(); } just returns "the last thing that was in the rax register"

16:18

but it's more complicated than that

16:18

but because it's hard to observe it, it's hard to understand it, I guess

you just need to have an example of every ub

16:19

examples

16:19

so students can analyze it and make conclusions

if you are suggesting "let them try and see what happens", then they might get to the wrong conclusion

16:20

e.g. that int *p = nullptr; *p = 0; always crashes

would u feel bad for doing X thing that a coworker was trying to fix because you knew how to fix it right away? i tried to explain how to do the fix but i kinda sucked explaining xD

16:25

and he seemed kinda lost

16:25

but its on my mind now

16:25

im a bad person maybe

16:26

(this guy is rly competent, thats probs not the issue, the thing is he just entered the project)

16:26

i also wanted this fixed asap xd

heinrich5991

with rust it seems to be a little bit better

I wish there was a UBless C that would be easier to teach on. I would teach Rust but the problem is I'd have to teach with an arbitrarily crippled version of Rust to achieve my teaching goals, but that'd be a disservice to the student as it'd be bad Rust

Ryozuki

im a bad person maybe

Doubt that has anything to do with it, if he is competent and cares about it he will review it on his own time and figure out how you fixed it

16:29

Or he will ask followup questions if you didn't appear as a complete dickhead

ehh

16:29

our work relationship is awkward, we both introverts

16:29

altho we differ as devs

16:29

as far as i know he has 0 open source projects :p

16:29

but i know he is good

How do you know he is good if he is new and doesn't have public projects?

cuz what he has done in internal projects

16:30

in the company

16:30

well

16:30

so far looks like he knows

16:31

altho sometimes idk

16:31

he sometimes said "you probs need unsafe to do X" but i pointed out with refcells and interior mutability u dont need unsafe xd

16:31

things like this

16:32

definitly more than ur average web dev

16:32

his uni project was doing simd matrices in assembly or smth i forgot tbh

Ryozuki

would u feel bad for doing X thing that a coworker was trying to fix because you knew how to fix it right away? i tried to explain how to do the fix but i kinda sucked explaining xD

overall, if he didn't look sad/angry I wouldn't care either

imagine a constant expresionless person

16:33

idk lmao

16:33

anyway moving on

he doesnt seem to care

16:35

Learath2

Programming especially benefits from practice imo. The more you code the better you get

+1

16:37

im just reading the convo

gerdoe

c is just good language

learning language*

Voxel

so all i need to do is to rename Single to KillMsg and then that's it? Idk what i'm supposed to be moving where @fokkonaut

was this question ever answered

gerdoe

but borrow checker uh

borrow checker good

i wanna know what im doing when i get home from .work

gerdoe

its literally unexpected behavior xd

do you know what is ub and why happens?

Voxel

was this question ever answered

i dont think he answered

I don't think that's what he means

16:44

he wants a KillMessage parameter that functions differently from the current one

but why

the current parameter name Single doesn't really capture what it does

17:34

I'm not entirely sure what he proposes or why

17:35

this thing should go btw: MsgPlus.m_Size = Teams()->TeamSize(Team());

17:35

this is information the client has

17:35

but in general, only very few things are still missing

17:35

or should be changed

17:36

it looks quite good already

heinrich5991

the current parameter name Single doesn't really capture what it does

I'll be renaming that (edited)

@Voxel you can get rid of that parameter if you change the code inside CCharacter::Die to something along these lines

bool TeamKill = Team() != 0 && Teams()->GetTeamState(Team()) == CGameTeams::TEAMSTATE_STARTED;

if(TeamKill)
{
    // Send your team kill message here
}
else
{
    // send the kill message
    CNetMsg_Sv_KillMsg Msg;
    Msg.m_Killer = Killer;
    Msg.m_Victim = m_pPlayer->GetCID();
    Msg.m_Weapon = Weapon;
    Msg.m_ModeSpecial = ModeSpecial;
    Server()->SendPackMsg(&Msg, MSGFLAG_VITAL, -1);
}

OHHH

18:35

waot

if the tee is not in a team you send single kill message, if he's in a team that is locked or whatever u send team kill message

@heinrich5991 do you remember how you added the hungarian notation? i cannot figure out how to not add prefixes for primitive types. i tried these but none seem to work.

- key: readability-identifier-naming.HungarianNotation.PrimitiveType.int
  value: null
  value:
  value: none
  value: None

I guess it should also be possible using the normal options, but this would be nicer. (edited)

Ravie

@Voxel you can get rid of that parameter if you change the code inside CCharacter::Die to something along these lines

bool TeamKill = Team() != 0 && Teams()->GetTeamState(Team()) == CGameTeams::TEAMSTATE_STARTED;

if(TeamKill)
{
    // Send your team kill message here
}
else
{
    // send the kill message
    CNetMsg_Sv_KillMsg Msg;
    Msg.m_Killer = Killer;
    Msg.m_Victim = m_pPlayer->GetCID();
    Msg.m_Weapon = Weapon;
    Msg.m_ModeSpecial = ModeSpecial;
    Server()->SendPackMsg(&Msg, MSGFLAG_VITAL, -1);
}

wait, you mean the Single parameter?

18:56

if I remove that, it will send the kill message for every member in the team

ah I c, but you can still create the TeamKill bool inside rather than an additional parameter

Isn't that what I did?

nah you added this bool Single

19:08

you can see inside the function itself if it's a single kill or team kill, no need to pass some bool

if I remove that, it will send the kill message for every member in the team

19:09

Teams()->OnCharacterDeath(GetPlayer()->GetCID(), Weapon);
}

19:09

it iterates through all the IDs and kills each tee. i need a way to prevent duplicates

but you can let it kill the tees normally and inside CCharacter::Die you decide if it should send the message

but how does it know if it's the one killing?

19:14

like, how does it detect who's strong

it doesn't know that anyway

19:17

the only difference from how you have it currently is that you do the check inside the function rather than passing a bool

there is a guaranteed order of spawn iirc though

19:26

the person who kills will have the lowest entity id iirc

Kotlin as a first OOP language? :p

marmare_314

@heinrich5991 do you remember how you added the hungarian notation? i cannot figure out how to not add prefixes for primitive types. i tried these but none seem to work.

- key: readability-identifier-naming.HungarianNotation.PrimitiveType.int
  value: null
  value:
  value: none
  value: None

I guess it should also be possible using the normal options, but this would be nicer. (edited)

I made my own script on top of libclang, didn't use clang-tidy, sorry

https://news.ycombinator.com/item?id=35330438

taubek

Employees Are Feeding Sensitive Biz Data to ChatGPT, Raising Security Fears

Ryozuki

https://news.ycombinator.com/item?id=35330438

not just at private companies either, I've heard very scary things from people that work at government jobs

@marmare_314 have you tried value: ""

20:12

?

heinrich5991

@marmare_314 have you tried value: ""

yes, and also value: '' which is supposed to work for the normal prefix options

what does it do if you try "" or ''?

heinrich5991

they think int function() { g(); } just returns "the last thing that was in the rax register"

imagine having a rax register

on amd64

Learath2

I wish there was a UBless C that would be easier to teach on. I would teach Rust but the problem is I'd have to teach with an arbitrarily crippled version of Rust to achieve my teaching goals, but that'd be a disservice to the student as it'd be bad Rust

not completely UBless, but -pedantic -Wall -Wextra -Werror gives you way less error already

heinrich5991

on amd64

yeah, but C is platform agnostic

heinrich deletor

20:59

Its Monday

20:59

heinrich is like the elderly lady not allowing kids to play on the grassfield

Chairn

yeah, but C is platform agnostic

the C standard doesn't talk a lot about platforms, it's not completely agnostic though. the "not completely agnostic" isn't relevant to the the current discussion though

21:00

but I think my point still stands, even if you add "on current amd64 ABIs" to the statement

21:02

see also https://utcc.utoronto.ca/~cks/space/blog/programming/CAsAbstractMachine on the point of "C is just an abstract machine"

heinrich5991

e.g. that int *p = nullptr; *p = 0; always crashes

doesn't crash on lots of embedded system

yes, it also doesn't crash cat (edited)

21:02

I was talking about amd64, though

cat?

yes

what's cat here?

$ echo 'int *p = nullptr; *p = 0;' | cat
int *p = nullptr; *p = 0;

21:04

didn't crash for me

21:05

what I was trying to say was that you're talking about microcontrollers when I already specified that I'm talking about amd64

21:05

i.e. not really relevant for the point I was trying to make

21:05

I was trying to say that instructors say something like int *p = nullptr; *p = 0; always crashes on amd64

21:05

which isn't true

21:06

trying to point out that some instructors in university don't understand C, even if you try to explain what UB is

heinrich5991

what does it do if you try "" or ''?

it ignores the value, probably related to https://github.com/llvm/llvm-project/issues/56358

readability-identifier-naming does not allow overriding a prefix wi...

I wish to enforce a particular prefix on global variables which are not constants. I would expect to be able to do this via the following options to readability-identifier-naming: - key: readabilit...

heinrich5991

trying to point out that some instructors in university don't understand C, even if you try to explain what UB is

or they just simplify it to not complicate things too much

but when asked about that point, they should be able to tell you "I just simplified it" and not "you're wrong, check this sample program", right?

marmare_314

it ignores the value, probably related to https://github.com/llvm/llvm-project/issues/56358

i just noticed, the exact issue is linked there so i guess i'll just try the other option

Ryozuki

do you know what is ub and why happens?

ub for me is absence of guarantee from language standard or what language comes up with

21:49

i can actually call possibility of us states to leave usa in xix ub just because constitution had nothing about states right of leaving

21:50

constitution nor allowed this nor prohibited it

ReD

Kotlin as a first OOP language? :p

Isn't it basically java with a new syntax? I mean u probably learn enough about oop to understand the idea behind it. But maybe too few to understand it's flaws^^